Re: "all" not inclusive of "replication" in pg_hba.conf

On Mon, 2011-09-26 at 14:13 -0500, Merlin Moncure wrote:
> On Mon, Sep 26, 2011 at 1:47 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
> > On Mon, 2011-09-26 at 11:58 -0500, Merlin Moncure wrote:
> >> On Sat, Sep 24, 2011 at 10:07 PM, Rajesh Kumar Mallah
> >> <mallah(dot)rajesh(at)gmail(dot)com> wrote:
> >> > Dear List ,
> >> >
> >> > It is been found that the entry
> >> >
> >> > local all all trust
> >> >
> >> > does not renders below redundant in pg_hba.conf
> >> >
> >> > local replication replicator01 trust
> >>
> >> I noticed this too, and I think it should. Either way, the
> >> documentation isn't clear on this point -- either 'all' should include
> >> the faux replication database or it should be noted in appropriate
> >> places that 'all' doesn't/can't do that.
> >>
> >
> > "all" includes all real databases, not "virtual" one. The documentation
> > could probably be clearer, but "all" shouldn't include the virtual
> > "replication" database.
>
> ok, what's your rationale for that? pg_hba.conf is a rule based system
> with no distinction given for rule vs virtual databases. what if we
> create more virtual databases? do you always have explicitly create a
> rule for each database for each user? IMSNHO, the more I think about
> it, the more I think current behavior is broken.
>

And I would agree (that the current behaviour is broken). Using a
database name as a flag to replication connection was a false good idea.
But, actually, I failed to find a better one.

--
Guillaume
http://blog.guillaume.lelarge.info
http://www.dalibo.com